Protein binding domains can be predicted from sequence data, however re-designing proteins with improved or altered binding affinities often requires testing of a number of variants of the re-designed protein. Currently the best method for obtaining proteins with desired binding affinities is to generate and screen a protein library including such variants that can include rationally redesigned proteins, randomly altered proteins, or a combination thereof. Libraries of many types of protein, such as immunoglobulins and scaffold proteins and receptors or receptor ligands have successfully been constructed and screened for binding affinity.
There are many methods to screen libraries, but one of the most common methods is the phage display method, which comprises fusion of a protein library to the coat proteins of filamentous phage (e.g., Huse et al., '89; Clackson et al., '91; Marks et al., '92). Fusions are made most commonly to a minor coat protein, called the gene III protein (pill), which is present in three to five copies at the tip of the phage. The fused library is then displayed on the surface of the viral particle. The page display library can then be screened against an immobilized target protein. However, one major drawback of this method is that target proteins that bind the library with very high affinity are not always identified because the conditions required to elute the bound phase usually denature the phage particle such that it becomes impossible to identify the protein of interest. Another draw back of phage display libraries is the requirement that the target protein be immobilized on a solid surface, which can lead to difficulties in determining the actual affinity of a target protein for the phage display protein. Furthermore, some proteins of interest require post-translational modifications, such as glycosylation, methylation, or disulfide binding, that cannot be achieved when expressed in a phage particle.
An alternative method for screening protein libraries is to display the library on the surface of bacterial cells. This method solves many of the drawbacks associated with phage display, but has its own problems. One problem with bacterial display is that the bacterial capsule can cause steric hindrance to proteins displayed on the bacterial surface. Also, bacteria do not contain the machinery to properly fold eukaryotic proteins, so the protein of interest may not always be expressed within the bacterium. Similar to the problem in phage, bacteria cannot provide post-translational modifications, like disulfide binding, to a eukaryotic protein.
Wittrup et al. (U.S. Pat. Nos. 6,699,658 and 6,696,251) have developed a method for a yeast cell display library. This is a two component system, wherein the first component involves expressing one subunit of the yeast mating adhesion protein, agglutinin, which is anchored to the yeast cell wall. The second component involves expressing a protein library fused to a second subunit of the agglutinin protein which forms high affinity disulfide bonds to the first agglutinin subunit. The protein library fused to the agglutinin is thus displayed on the surface of the cell. The library can then be screened. This method allows for the proper folding and post-translational modification of eukaryotic proteins.
Rakestraw et al. (PCT/US2008/003978) have developed a three component system for displaying a protein library on the surface of yeast cells. The first component involves expressing a protein library fused to a biotin-binding peptide, the second component involves modifying the yeast cell wall to express biotin, and the third component involves binding avidin to the biotin expressed on the cell surface. The fused protein library is then biotinylated and secreted from the yeast cell and binds to the avidin on the yeast cell surface, thus displaying the protein library on the surface of the yeast cell. One potential drawback of this system is that avidin non-specifically binds all biotin. Another potential drawback is that avidin contains four binding sites, which may cause steric hindrance thus preventing the biotinylated protein library from binding to the cell surface bound avidin. Similarly, the avidin molecule may bind the biotinylated protein library before binding the biotinylated yeast cell wall, thereby hindering the binding of the avidin to the yeast cell wall. Additionally, this method contains the added complication of having to biotinylate the protein library within the yeast cell. This necessary extra step requires further modification to the yeast cell. It is well established that the more modifications are made to a biological system, the less likely it is that the system will behave as designed. In addition, since avidin/streptavidin is multivalent, care must be taken to not cross-link the biotinylated cells. Finally, biotin/streptavidin and biotin/avidin are used in a number of commercially available labeling kits. Such kits would be difficult to use in such a biotin/avidin display system.
The prior art lacks a simple, efficient system capable of specifically binding a secreted protein library using an adapter molecule that binds to the protein library and to the surface of a eukaryotic cell through different binding moieties.